Peter Jones wrote:
"I'm just wondering whether _for me_ the pattern of thought is much more
Question-Argument-Idea...etc.? And that that proceeds left-to-right
visually
and logically.
Ah, OK. A good point, that also points back to the discussion about whether
it's better to have semantics in links.
Peter points to an apparent ambiguity in the QM (and Mifflin) GUI -- the
link arrows point "backwards" Actually the discourse does generally proceed
exactly as Peter describes -- questions, then possible answers, then
arguments (or further questions).
Nearly everyone (including myself, far back in the deeps of time) has
trouble with this when they first encounter it.
Jeff Conklin would be a better one to talk about why QM was done this way,
but I'll give my take on it.
The link arrow, generally speaking, means that the node on the right is in
some way "about" the one on the left that it points to. So, for example, if
we have:
"What is the best GUI for collective discourse?" <--- "Mifflin and QM"
...this means that the node "Mifflin and QM" is about, or addresses, the
question "What is the best GUI for collective discourse?".
Although this does indeed seem back to front at first, in practice we've
found that there is actually a greater clarity to this usage. Arrows
pointing the other way (i.e. left to right) turn out to be kind of
ambiguous. They mean different things in different contexts, but the
meaning is only implicit -- and thus open to misinterpretation. (Similarly,
many "flowcharts" are drawn where boxes and arrows are used to mean
different and inconsistent things -- the meanings may be obvious to the
person drawing it, and those who saw it being drawn, but will not be to
later people looking at the picture).
Thus in QM and Mifflin, arrows always mean the same thing and do not "hide"
further implicit meanings. In fact, one of the core contributing
disciplines to Compendium was the CommonKADS knowledge modeling
methodology, which always prescribed precise and unvarying meanings for
each symbol used in any diagram. When I "got" this message, finally, after
long resistance, I saw that such structures, far from restricting freedom
of expression, actually contribute to clarity and focus of expression,
which is at the core of Compendium.
This is also why I don't so much like semantics in links -- it complicates
the understanding of what they mean (although I accept Simon's arguments on
this score). The real issue is coming up with the 'right' ways to indicate
how something relates to something else. In Compendium we encourage
questions to be used for this purpose. For example, let's say we have two
ideas:
"Great GUIs allow flexible manipulation of concepts"
and
"Mifflin and QM's GUI"
We could indicate the relationship between these two as follows:
"Great GUIs allow flexible manipulation of concepts" <--"example of"--
"Mifflin and QM's GUI"
where "example of" is a link.
Compendium would render this as follows:
"Great GUIs allow flexible manipulation of concepts" <--"Examples?" <--
"Mifflin and QM's GUI"
where "Examples?" is a Question node, not a link.
Why is this "better"? Well, I would argue that it's because posing it as a
question allows multiple possible answers, each with its own possibility to
spawn further discussion and inquiry. Each piece (subset of nodes and
links) of the collection can be pulled off by itself for closer focus.
So the above example could lead to:
"Great GUIs allow flexible manipulation of concepts" <--"Examples?" <--
"Mifflin and QM's GUI"
<-- "Nodal's GUI"
<-- "Peter Jones's GUI"
(hopefully the above looks like all the "GUI" ideas are linked to the
"Examples?" question)
Subsequent argumentation and/or discussion of each of the possible answers
can clearly and explicitly address whether each of the GUIs are indeed good
Examples of Great GUIs. It's clear and explicit because they're all linked
to the same Question.
If, instead, this was done with separate semantic links:
"Great GUIs allow flexible manipulation of concepts" <--"example of"--
"Mifflin and QM's GUI"
<--"example of"-- "Nodal's GUI"
<--"example of"-- "Peter Jones's GUI"
....then subsequent argumentation or discussion will be only indirectly
related to the original concept, and the question that the subsequent
discussion is addressing will be 'buried' -- it requires participants to
look at each individual link, determine its relationship to the original
concept, etc.
All of this, I guess, is part of the rhetorical goal to get people to think
in terms of questions and possible answers, in the service of clear and
generative communication. This approach doesn't reduce the logical clarity
or computability of discussions so rendered, either.
By the way, one of the ways in which Compendium as a methodology diverged
from 'classic' IBIS, is that we downplay even the rhetorical link types
found in QuestMap (Expands On, Specializes, etc.). We found that requiring
users to choose such link types stymied almost all novices, especially when
in front of a group watching what they were doing, without (usually) adding
a huge amount of value. Mifflin (although it supports the QM link types)
relies on a single generic link type, which means 'about' or 'addresses'.
Al
------------------------ Yahoo! Groups Sponsor ---------------------~-->
E-mail viral marketing - with FastTree
Email to 50. You might reach 500.
Unlimited use and tracking, $20/mo. http://www.fasttree.com/s/11.htm
http://us.click.yahoo.com/7XXUJB/MJRDAA/fzPGAA/IHFolB/TM
---------------------------------------------------------------------~->
Community email addresses:
Post message: unrev-II@onelist.com
Subscribe: unrev-II-subscribe@onelist.com
Unsubscribe: unrev-II-unsubscribe@onelist.com
List owner: unrev-II-owner@onelist.com
Shortcut URL to this page:
http://www.onelist.com/community/unrev-II
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
This archive was generated by hypermail 2.0.0 : Fri Nov 09 2001 - 03:12:25 PST